Going into election night 2004, I took the nine states that were popularly considered up for grabs — Michigan, Minnesota, Ohio, Pennsylvania, Wisconsin, Florida, Iowa, New Mexico, and Nevada— and assumed the others would go the same way they did in 2000. (My "safe" states were all correct, except for New Hampshire whose four electoral votes switched from Bush to Kerry.) That meant there were 512 possible electoral vote scenarios and, because so many of the up for grabs electoral votes were in states that Gore won in 2000 (Kerry was defending 70 of Gore's precious electoral votes, while Bush had to defend just 52 he had previously won in order to be reelected), my odds showed Bush was a 70 percent favorite to win reelection, with 358 scenarios in which he won compared to Kerry's 143. (Had I added New Hampshire to my odds charts, Bush would have still won in 67 percent of the possible outcomes.) You could focus on the narrow chances Kerry had to win in 2004 — that is, he would have been President if just 59,388 Ohio votes switched from Bush to Kerry — but going into election night, it would have been one of the unlikeliest outcomes. Anyone who's lost on a bad beat in poker knows it's possible for someone to pull a fourth jack on the river against your flush, but it's not a good bet.
In 2008, the odds shifted strongly in the Democrats' favor, largely because, whether through the charms of Barack Obama, demographic changes in the intervening four years, the worst financial panic since the Great Depression, or an inept McCain campaign, there were more states that had previously voted Republican that were in play for the Democrats. In addition to the nine states that were up for grabs in 2004 there was also a plausible case that Obama could repeat Kerry's victory in New Hampshire and pick up two states that had voted for Bush in the two prior elections, Colorado and Virginia. With those 11 states in play, my odds chart showed Obama winning 77 percent of the 2,048 possible electoral vote scenarios. As it turned out, I was too conservative in estimating Obama's odds, as he won ten of those states as well as the electoral votes from places no one thought were in play: Indiana, North Carolina and the 2nd District of Nebraska.
At a cocktail party a few years back, I described my odds chart to Silver, who goodnaturedly scoffed at its lack of sophistication. He's right, of course: He and other stat-savvy data-crunchers are interested in predicting, as precisely as possible, how people will vote. I come from the other direction: I assume I can't know how the vote will turn out, so it's better to look at the entire universe of possible outcomes. And in this election, the math says Romney has lots of ways to win. Just not nearly as many as Obama.
This article is from the archive of our partner The Wire.